Download Deciphering the Sox-Oct partner code by quantitative cooperativity

Published online 16 February 2012
Nucleic Acids Research, 2012, Vol. 40, No. 11 4933–4941
doi:10.1093/nar/gks153
Deciphering the Sox-Oct partner code by
quantitative cooperativity measurements
Calista K. L. Ng1,2, Noel X. Li1, Sheena Chee1, Shyam Prabhakar3,
Prasanna R. Kolatkar1,4 and Ralf Jauch1,*
1
Laboratory for Structural Biochemistry, Genome Institute of Singapore, 60 Biopolis Street, Singapore 138672,
School of Biological Sciences, Nanyang Technological University, Singapore 637551, 3Computational and
Systems Biology, Genome Institute of Singapore, 60 Biopolis Street, Singapore 138672 and 4Department of
Biological Sciences, National University of Singapore, Singapore 117543
2
Received October 4, 2011; Revised and Accepted January 25, 2012
ABSTRACT
Several Sox-Oct transcription factor (TF) combinations have been shown to cooperate on diverse
enhancers to determine cell fates. Here, we developed a method to quantify biochemically the
Sox-Oct cooperation and assessed the pairing of
the high-mobility group (HMG) domains of 11 Sox
TFs with Oct4 on a series of composite DNA
elements. This way, we clustered Sox proteins according to their dimerization preferences illustrating
that Sox HMG domains evolved different propensities
to cooperate with Oct4. Sox2, Sox14, Sox21 and
Sox15 strongly cooperate on the canonical element
but compete with Oct4 on a recently discovered compressed element. Sry also cooperates on the canonical element but binds additively to the compressed
element. In contrast, Sox17 and Sox4 cooperate more
strongly on the compressed than on the canonical
element. Sox5 and Sox18 show some cooperation
on both elements, whereas Sox8 and Sox9 compete
on both elements. Testing rationally mutated Sox
proteins combined with structural modeling highlights critical amino acids for differential Sox-Oct4
partnerships
and
demonstrates
that
the
cooperativity correlates with the efficiency in
producing induced pluripotent stem cells. Our
results suggest selective Sox-Oct partnerships in
genome regulation and provide a toolset to study
protein cooperation on DNA.
INTRODUCTION
How regulatory information is genetically encoded is an
overarching yet unresolved question in genome biology.
This information is scanned and interpreted by
sequence-specific transcription factor (TF) proteins.
However, the biochemical basis for the selective recruitment of TFs to genomic enhancers that govern spatial and
temporal gene expression remains elusive. Multiple studies
have shown that TFs often bind to short and degenerate
DNA-binding sites that have been discovered computationally in huge numbers throughout the genome (1–3).
Yet, only 1–5% of these binding sites are actually
occupied by the corresponding TF. How do TFs discriminate between functional and nonfunctional binding sites?
It has been shown that TFs have a propensity to cluster
and are more likely to target genomic regions that are
co-bound by other factors (4,5). Potentially, enhancers
of co-expressed genes could share their own distinctive
‘fingerprint’ or grammar of DNA motifs that recruit particular TF combinations. To predict gene expression
patterns from DNA sequence and TF concentration
alone, this motif grammar needs to be decoded. It is
possible that enhancers of co-expressed genes are only
loosely defined with an unconstrained arrangement of
binding motifs over several 100 bps not necessitating the
direct physical interactions of TFs (4,6). In contrast, the
motif grammar may include binding sites with constrained
spacing between them whose recognition is tied to specific
protein–interaction surfaces of individual TF proteins.
These protein interactions underlie their developmental
specificities and selectively target TFs to genomic enhancers. However, while TF heterodimerization predominates
among paralogous groups of TFs such as nuclear receptors (7), helix–loop–helix (8) and leucine zipper families
(9), examples for the selective dimerization of structurally
unrelated TFs are sparse. Nevertheless, several studies
have highlighted the importance of a direct cooperation
between unrelated TF pairs (10–13). Most prominently,
the Sox and Oct families of TFs have been shown to cooperate to execute key developmental programs (14,15).
*To whom correspondence should be addressed. Tel: +65 6808 8102; Fax: +65 6478 9060; Email: [email protected]
ß The Author(s) 2012. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/3.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
4934 Nucleic Acids Research, 2012, Vol. 40, No. 11
On their own, all 20 mammalian Sox proteins bind to a
CTTTGT-like sequence (2,16,17) while most Oct factors
recognize an octamer related to a ATGCAAAT sequence
(18). By combining these sequences, composite motifs can
be constructed with different motif orientation and
spacing (19). Several such composite motifs have been
found to be functional targets for the synergistic regulation by Sox and Oct proteins. For example, (i) the
Sox2-Oct4 pair drives stem-cell pluripotency genes on
either a 0 or a 3 bp-spaced motif element (20–23); (ii)
Sox17-Oct4 cooperate during endodermal differentiation
(24) presumably on a compressed motif (19) and (iii)
Sox2-Brn2 regulates brain development on a sox-oct
motif with a 6-bp spacer (13). Notably, when the cooperative binding of Sox2 and Oct4 to DNA is perturbed by
rational mutagenesis its ability to induce pluripotency is
lost (19). Conversely, although wild-type Sox17 cannot
induce pluripotency, a mutated version of Sox17 increases
cooperative binding with Oct4 on pluripotency gene enhancers and thus has the potential to induce pluripotency.
Such results suggested that there might be a Sox-Oct
partner code that underlies cell fate decisions (14,15). To
further investigate whether members of the Sox and Oct
families evolved features to cooperatively target specific
enhancer elements a global assessment of the Sox-Oct
pairing profile is highly desirable. To this end, we have
developed a method to measure heterodimer cooperativity
factors revealing the mode TF heterodimerization on composite DNA elements. We used this method to study the
heterodimerization
propensities
of
representative
members of all seven major Sox families with Oct4 on a
range of composite DNA elements. As a result, we found
that Sox families exhibit markedly different propensities
to associate with Oct4 on distinctly configured binding
motifs. By measuring cooperativity factors of rationally
mutated Sox proteins, we found that the re-engineered
Sox17EK behaves like an enhanced Sox2. This likely
underlies its improved properties in producing induced
pluripotent stem cells in lieu of the native pluripotency
factor Sox2 (19). Together, we demonstrate that
cooperativity measurements are critical to understand
TF function and the cis-regulatory logic of developmental
enhancers.
MATERIALS AND METHODS
Cloning, protein expression and purification
The POU domains of mouse Oct4 and high-mobility group
(HMG) domains of mouse Sry, Sox2, 14, 21, 4, 5, 8, 9, 17, 18
and 15 were BP cloned from their respective Imagene clones
(Oct4: IRAKp961K04111Q; Sry: IRAMp995I2211Q;
Sox2: RPCIB731A06406Q; Sox14: IRAKp961K05125Q;
Sox21: IRAKp961C14126Q; Sox4: IRAVp968F02163D;
Sox5: IRAMp995N0310Q; Sox8: IRAVp968H01144D;
Sox9: IRAVp968B0369D; Sox17: OCACo5052D058;
Sox18: IRAVp968E0317D; Sox15: IRCKp5014B242Q)
into a pENTRY vector, pDONR221 (see Supplementary
Table S1C for primer sequences). The resulting
pENTR-constructs were first verified by sequencing
and recombined into either pETG20A or pHisMBP
expression plasmids using the GATEWAYTM technology
(Invitrogen). Constructs were transformed into Escherichia
coli BL21(DE3) cells, grown in 1 Terrific Bertani broth
supplemented with 0.1% glucose and 100 mg/ml ampicillin
until OD600 nm 0.6–0.8 before inducing with 0.5 mM
isopropyl-b-thiogalactoside at 18 C for 18 h. Fusion
proteins were purified using previously published protocol
(19,25,26). In short, fusion proteins were purified using an
immobilized metal affinity chromatography step, tag
cleavage using the TEV protease followed by ion-exchange
chromatography and gel filtration.
Electrophoretic mobility shift assays
Electrophoretic mobility shift assays (EMSAs) were
carried out using forward strand 50 Cy5-labeled-dsDNA
(Sigma Proligo, see Figures 2A and 3A). DNA probes
were prepared by mixing equimolar amounts of complementary strands in 1 annealing buffer (20 mM Tris–HCl,
pH 8.0; 50 mM MgCl2; 50 mM KCl), heated to 95 C for
5 min and subsequently with 1 C/min ramping down to
4 C in a PCR block. Typical binding reactions contain
100 nM dsDNA with varying concentrations of both Sox
and Oct4 proteins in a 1 EMSA buffer [20 mM Tris–
HCl pH 8.0, 0.1 mg/ml bovine serum albumin (Biorad),
50 mM ZnCl2, 100 mM KCl, 10% (v/v) glycerol, 0.1%
(v/v) Igepal CA630 and 2 mM b-mercaptoethanol] and
were incubated for 1 hr at 4 C in the dark to reach
binding equilibrium. Reactions were loaded into a
pre-run 12% native 1 Tris–glycine (25 mM Tris pH
8.3; 192 mM glycine) polyacrylamide gel, and DNA
complexes were separated at 4 C for 30 min at 200 V.
The bands were detected using a Typhoon 9140
PhosphorImager (Amersham Biosciences) and quantified
using the ImageQuant TL software (Amersham
Biosciences).
Cooperativity factor measurement
As an extension of our homodimer model described previously (27), we defined four possible microstates for a
heterodimer-binding model. The participating species
are defined as D for DNA, P1 for protein 1 and P2 for
protein 2. The equilibrium dissociation constant of each
individual protein can be represented as in Equation 1
where [D] is the concentration of free DNA, [P1] and
[P2] concentrations of free proteins and [DP1] and [DP2]
solitary protein–DNA complexes.
Kd1 ¼
½D½P1
½DP1
or
Kd2 ¼
½D½P2
½DP2
ð1Þ
If P1 and P2 are mixed with DNA in the same tube, the
fourth state representing a ternary complex becomes
feasible. Dissociation constants of secondary-binding
events are described by Equation 2, where [DP1P2] is
the equilibrium concentration of heterodimeric protein–
DNA complexes.
Kd12 ¼
½DP1½P2
½DP1P2
and Kd21 ¼
½DP2½P1
½DP1P2
ð2Þ
Nucleic Acids Research, 2012, Vol. 40, No. 11 4935
With f0, f1, f2 and f3 defined as the fractional concentrations of the free DNA, monomer–DNA complexes 1 and 2
and the heterodimer–DNA complex, respectively
(f0+f1+f2+f3 = 1), the heterodimer cooperativity factor
! can be straightforwardly calculated from the experimentally determined fractional concentrations f0, f1, f2 and f3
as follows:
!¼
Kd1
Kd2
½D½DP1P2 f0 f3
¼
¼
¼
Kd21 Kd12 ½DP1½DP2 f1 f2
ð3Þ
As defined here, ! > 1 implies positive cooperativity;
! = 1 no cooperativity; ! < 1 negative cooperativity. To
reduce errors when calculating !, we only included measurements, where each of the four fractional concentrations
were at least 5%. Cooperativity factor heatmaps and
graphs for the Sox-Oct4-DNA combinations were
plotted using R (http://www.r-project.org/) with the
Gplot package.
Structural modeling and analysis
Homology models for Sox HMG or the Oct4 POU
domain proteins were generated using I-TASSER with
the Sox17 HMG (pdb-id 3F27) and the Oct1 POU
(pdb-id 1GT0) as templates (http://zhanglab.ccmb.med.
umich.edu/I-TASSER/) (28). Sox HMGs were subsequently superimposed onto the Sox2 HMG/Oct1 POU
complex bound to a canonical element (pdb-id 1O4X).
Superpositions, visual inspections and figure generation
were carried out using PyMol.
A method for the determination of cooperativity factors
We first cloned HMG domains of mouse Sox proteins and
screened for soluble protein expression in a 96-well
format. Representative members of most Sox families
were then purified to at least 95% purity. Next, we
performed EMSAs to quantify the fractional binding of
Sox proteins and Oct4 on DNA. At equilibrium conditions, the mixing of the two proteins with DNA results
in macromolecular complexes corresponding to four
microstates per EMSA lane: (1) free DNA; (2)
Sox-bound DNA; (3) Oct4-bound DNA and (4)
Sox-Oct4-DNA ternary complex (Figure 1C). The abundance of each microstate at equilibrium is directly proportional to its Boltzmann weight, which in turn is a function
of the protein concentrations, the equilibrium dissociation
constants and the cooperativity factor, ! (Figure 1D, see
‘Materials and Methods’ section). Substituting the fractional contribution of each microstate from equilibrium
experiments into our heterodimerization cooperativity
model allowed us to quantify the cooperativity factor
(Figure 1D). The cooperativity factor is essentially the
fold change in the equilibrium binding constants when a
protein co-binds, relative to the equilibrium constant for
solitary binding. Values greater than 1 represent positive
cooperativity, where both proteins mutually lower their
free energies of binding. That is, complex formation is
favoured. Negative cooperativity (! < 1) represents a
competitive-binding mode, where the protein has a preference for binding to unbound DNA rather than forming
a ternary complex. Finally, values of about 1 correspond
to additive binding with proteins having no specific preference to binding either DNA that was already bound by
another protein or free DNA.
RESULTS AND DISCUSSION
Sox proteins exhibit diverse protein interaction surfaces
Sox proteins exhibit a unique dimerization preference
with Oct4 on variant DNA configurations
The 80 amino acid HMG domain of Sox proteins is highly
conserved for all paralogs (Figure 1A). In accordance with
similar DNA sequence preferences for all Sox proteins (2),
amino acids that contact DNA bases are nearly invariant
for all Sox proteins. However, protein contact interfaces
as defined in structural studies on Sox2 and Oct1 show
some disparity (30) (highlighted as blue empty circles). As
an extension of earlier work on Sox2 and Sox17, we
generated homology models for all Sox families and inspected the electrostatic charge distribution on the van der
Waals’ surface (Figure 1B). The protein surface of Sox
proteins facing Oct4 when bound to canonical sox-oct
motifs show pronounced differences distinguishing Sox
families. The SoxC, E and F groups contain an acidic
patch at this interface, the SoxB and SoxG groups are
highly basic and the SoxD group is largely neutral. We
have recently shown that residue 57 (numbering according
to HMG conventions), which is causing the disparate electrostatic pattern of the Sox families, is critical for the effective dimerization of Sox2 with Oct4 on pluripotency
enhancers (19). To understand how these structural differences affect Sox-Oct partnerships, we developed a quantitative method to study TF cooperation and analysed the
interaction of 11 Sox proteins with Oct4.
We measured cooperativity factors in multiple replicates
for 11 Sox proteins with Oct4 on nine differently
configured composite DNA motifs including the ‘canonical’ Sox2-Oct4 site found in many embryonic stem cell
enhancers (23,31), the plus3 site found in the Fgf4
enhancer (32) and the newly discovered ‘compressed’
element (19). In particular, on the canonical and compressed elements, we observed differences in the
cooperativity pattern for the Sox proteins, whereas all
Sox proteins tested cooperated with Oct4 on the plus3
element (Figure 2B, C, E and Supplementary Figure S1).
We combined the whole dataset of log2 transformed
cooperativity factors and created a heat map using
the hierarchical clustering method implemented in the
heatmap.2 R package (Figure 2C). The clustering
approach revealed that the Sox proteins can be
categorized into five separate groups highlighting their
cooperativity patterns (Figure 2C). Similarly, DNA
motif configurations cluster into five groups. Cluster II
contains only the plus3 element displaying cooperative
recruitment of all Sox-Oct4 pairs. Widely spaced
elements (plus 2–plus 10; cluster III) exhibit an essentially additive-binding mode for all proteins under
study. The plus1 element, however, shows a strongly
4936 Nucleic Acids Research, 2012, Vol. 40, No. 11
Figure 1. (A) Alignment of amino acid sequence of all mouse Sox-high-mobility group (HMG) domains shaded with BOXSHADE. The Sox
subfamilies are indicated to the right. The numbering corresponds to the HMG convention (29). a-Helices are marked with a red bar. The
Phe-Met wedge is indicated with an orange bar below the alignment. DNA interacting residues are marked by black empty circles while Sox-Oct
interacting residues are marked by blue empty circles. Highly conserved and similar sequences are shaded in black or gray. (B) A phylogenetic tree
calculated using PROML (http://caps.ncbs.res.in/iws/proml.html). This simplified tree largely corresponds to the more exhaustive phylogenetic
analysis of Sox factors. Sox subgroups (29) and the amino acids found at position 57 of the HMG domains are indicated in single letter codes.
(continued)
Nucleic Acids Research, 2012, Vol. 40, No. 11 4937
competitive-binding mode for all Sox-Oct4. The canonical
element and the compressed element (clusters I and IV)
exhibit a strong disparity with regard to the Sox-Oct pairs
they preferentially recruit and determine the clustering of
Sox proteins. Sox8 and Sox9 (cluster B) are not capable of
cooperating with Oct4 on either the compressed or the
canonical element. By contrast, Sox5 and Sox18 (cluster
E) bivalently cooperate on both elements. Clusters A and
D, however, have inverse cooperativity profiles on those
elements. While Sox2, Sox15, Sox14 and Sox21 (cluster A)
strongly cooperate on the canonical element, they are not
capable of co-binding with Oct4 on the compressed
element. Similarly, Sry cooperates on the canonical
element but retains an additive-binding mode on the compressed element. By contrast, Sox4 and Sox17 (cluster D)
only weakly cooperate on the canonical element
but strongly cooperate on the compressed element.
Overall, while the HMG domains of the Sox proteins
investigated here bind highly similar DNA sequences in
isolation, they markedly differ in their potential to cooperate with Oct4.
Cooperativity patterns of rationally mutated Sox proteins
We noticed that the cooperativity-based classification of
Sox proteins shows some relationship with their evolutionary classification (Figure 1B). Residue 57, which was predicted to affect the cooperativity of Sox2 and Sox17 with
Oct4, provides a partial mechanistic explanation for this
result. A lysine (Lys, K) residue at position 57 appears to
favour cooperativity on canonical and plus3 configuration
but is not compatible with binding to the compressed
motif. We have previously shown that residue 57 is a
critical determinant of the developmental function of
Sox proteins (19). When this residue is swapped between
Sox2 and Sox17, that is the K is replaced by a glutamate
(Glu, E) and vice versa, their biological functions are
interchanged and Sox17EK turns into an inducer of
pluripotency and Sox2KE into a trigger of endodermal
differentiation. To quantify the effect these mutations
have on Oct4 cooperation, we compared the cooperativity
of the mutant HMG domains with their wild-type counterparts. For these experiments, we used an element
derived from the enhancer of the Nanog gene that
behaves similarly as the idealized sox-oct element in
cooperativity measurements (21). We found that the
cooperativity of the Sox17EK protein with Oct4 is
roughly 30 times stronger than that of wild-type Sox17
and even three times stronger than of wild-type Sox2
(Figure 3B and C).
Sox5 contains an alanine (Ala, A) at position 57 and
strongly cooperates on both the canonical and compressed
motifs. Given the pronounced effect of residue 57 on
Sox-Oct cooperativity, we asked whether the presence of
an A residue at position 57 in Sox5 might perhaps explain
its ability to bind Oct4 cooperatively on both the canonical and compressed motifs. Indeed, we found that the
Sox17EA mutation raises the cooperativity factor of
Sox17 on the canonical element 10 times more and
brings it up on par with Sox2 and Sox5 in binding the
canonical sequence (Figure 3B and C).
Next, we asked whether these amino acid swap mutations also interchange the dimerization propensities on the
compressed motif. As expected, the Sox17EK mutation
causes a 30-fold drop of cooperation compared to
wild-type Sox17 and now behaves like wild-type Sox2.
However, Sox2KE cooperates only marginally better
than wild-type Sox2 on this element indicating that
further modifications in Sox2 are required to engineer a
Sox17-like dimerization propensity on the compressed
element. Further, introducing the Sox5-like alanine into
Sox17 to generate the Sox17EA protein results in a
20-fold drop in the cooperativity although Sox5 cooperates strongly. We noted that Sox5 contains a glutamine
(Gln, Q) at position 56, which is unique for the SoxD
group (Figure 1A). All other Sox proteins contain an
alanine at this position. It is conceivable that Gln56
impacts the cooperation of Sox5 on the compressed
element by compensating for Glu57. The lack of both,
Gln56 and Glu57, could explain why Sox17EA and
Sox2KA cannot cooperate on the compressed element.
We were intrigued by the observation that Sox17EK
cooperates more strongly with Oct4 on the canonical
element than wild-type Sox2 (Figure 3B and C). We therefore decided to further explore this by allowing two different Sox proteins to compete for co-binding with Oct4 in
the same reaction tube (Figure 3D). Consistently,
Sox2-Oct4 complexes predominated when mixed with
Sox17, whereas the situation was inversed in the
presence of Sox17EK (Figure 3D). Interestingly, these
results correlated with our earlier observation that
Sox17EK produces induced pluripotent stem cells more
efficiently than Sox2 in reprogramming cocktails (19).
The enhanced ability of Sox17EK to cooperate with
Oct4 on pluripotency enhancers may thus be the basis
for this observation.
Figure 1. Continued
Electrostatic surface maps of representing Sox members were calculated as described (26). Positively and negatively charged regions were represented
in red and blue patches, respectively. Homology models for Sox HMGs were generated using I-TASSER (28) and surface patches that differ for Sox
groups are boxed. (C) Illustration of how the microstates of the DNA complexes were quantified using the ImageQuant TL software. The cy5-labeled
dsDNA migrated differently on native gel depending on how the proteins and DNA associate. Thus, the fractional contribution of the microstates of
the free DNA (f0), Sox-DNA (f1), Oct4-DNA (f2) and ternary complex (f3) can be quantified. (D) Schematic diagram highlighting the approach to
calculate the cooperativity of TF pairs on composite DNA elements. Boltzmann weights of the respective complexes are denoted as b_D, b_DP1,
b_DP2 and b_DP1P2 and scaled so that the b_D = 1. [P1] and [P2] are the concentrations of the free proteins. The cooperativity factor omega does
not depend on the concentration of the reactants but solely on the relative ratios of the four microstates represented by their fractional contributions
measured in (C) (see main text and alternate derivation of the equation in the ‘Materials and Methods’ section).
4938 Nucleic Acids Research, 2012, Vol. 40, No. 11
Figure 2. (A) Sequences of the idealized composite Sox-Oct-labeled probes used. The Sox-binding sites are indicated in orange while the Oct-binding
sites are indicated in blue; (B) Bar plots showing cumulative mean cooperativity factors for 11 Sox HMG domains for elements shown in (A). Raw
values and individual bar plots per element are shown in Supplementary Table S1 and Figure S1. To derive reliable omega values and to minimize
errors in band quantification, the concentration of Sox HMG and the Oct4 POU was adjusted, such that the fractional contribution of each of the
four microstates was at least 5%. If such conditions could not be established, that is, for maximally competitive binding excluding ternary complexes
as seen on the plus1 element for most Sox HMGs or Sox2-Oct4 pairing on the compressed element, omega values were set to 0.01. Constitutive
cooperativity was not observed in this study. (C) Heat map of cooperativity factors representing the different Sox-Oct4 dimers on the various DNA
motifs. Log2-transformed mean cooperativity factors are expressed in a three-color gradient: red (competitive), white (additive binding) and blue
(positive cooperativity). The matrix was hierarchically clustered using the heatmap.2 function in R with default parameters. Different categorizations
were labeled as Clusters A–E and I–V. Each cooperativity factor was derived from at least 3 and maximally 30 replicates (see Supplementary Table
S1). (D) Summary of the differential assembly dataset grouping Sox HMG domains exhibiting similar Oct4 cooperativity profiles. Candidate amino
acids that likely explain the disparate Oct4 interactions at positions 57 and 64 are shown. (E) Differential assemblies of different Sox HMG members
(50 nM) with the Oct4 POU protein (150 nM) were performed on compressed (left), canonical (center) and plus3 (right) element DNA. The cartoon
to the left symbolizes free DNA (black line), Sox (blue circles) and Oct (orange squares).
Structural determinants for Sox -Oct cooperation
Our findings show that residue 57 is critically important
for the discrimination between the canonical and
compressed motifs. To study the structural basis for the
differential assembly of Sox HMGs, we generated
homology models of several Sox HMG/Oct4 POU
complexes on the canonical element (Supplementary
Figure S2). We observed that K57 of Sox2 interacts with
a backbone carbonyl of the POU specific domain
(Supplementary Figure S2). When K57 is replaced by
Nucleic Acids Research, 2012, Vol. 40, No. 11 4939
A
B
C
D
Figure 3. (A) Sequences of the labeled Nanog element probes used (21). The Sox-binding sites are indicated in grey while the Oct-binding sites are
indicated as underlined; (B) Representative EMSAs of different Sox proteins with Oct4 on canonical and compressed motif. The indicate mutants
refer to amino acid position 57 of the HMG domain. (C) Cooperativity factors for various Sox mutants compared to their wild-type counterparts
expressed as mean ± standard deviation. (D) Competitive EMSA analysis of showing that the Sox17EK-Oct4 complexes predominate Sox2-Oct4
complexes, whereas the Sox2-Oct4 complex clearly outcompetes Sox17-Oct4 (lanes 9 and 10). A N-and C-terminally extended Sox2 HMG domain
(2L) comprising 109 amino acids (residues 33–141 of full length Sox2 protein) was used to distinguish the various complexes. The cartoon to the right
symbolizes free DNA (black line), Sox2L (grey-filled circles), Sox17 and Sox17EK (grey empty circles) and Oct4 (black squares).
E57, as in Sox17, Sox8 and Sox9, the negatively charged
carboxyl group of Glu likely causes unfavourable charge–
charge repulsions, leading to a drop in cooperativity. In
contrast, an A57, as found in Sox5 or in Sox17EA, is
compatible with binding.
However, residue 57 alone cannot explain the
cooperativity profiles of the whole Sox family and there
must be a combination of contributing elements. For
example, while the E57 proteins Sox17, Sox18 and Sox4
cooperate strongly on the compressed element, Sox8 and
Sox9 cannot, and Sox18 retains some cooperativity on the
canonical element. The structural modeling suggested
residue 64 as an additional candidate underlying the differential behaviour of Sox factors in our cooperativity
assays (Supplementary Figure S2). Residue 64 is placed
at the interaction interface and shows a strong divergence
within the Sox family (Figure 2D, Supplementary
Figure S2). Importantly, a M64E mutation has been
demonstrated to abrogate Sox2–Oct4 interaction (30).
Likewise, the charged K64 in Sox8 and Sox9 could
4940 Nucleic Acids Research, 2012, Vol. 40, No. 11
underlie their inability to interact with Oct4 on canonical
and compressed elements.
While exhibiting overall similar cooperativity patterns,
the degree of cooperativity still differs for Sox HMG
domains with identical residues at positions 57 and 64
(i.e. Sox14/Sox21 cooperate 5 times more strongly than
Sox2 on the canonical element). We hypothesize that those
residual difference are due to variations within the flexible
and only poorly conserved C-terminal tail of the HMG
domain that was shown to contribute to Sox2–Oct4 interactions (30) (Figure 1A). Experimental structures of
Sox-HMG-Oct4 combinations on canonical and compressed elements combined with mutagenesis experiments
are desirable to put those hypotheses to a test.
CONCLUSION
In this study, we have developed a biochemical system that
enables
quantitative
measurements
for
protein
heterodimerization on DNA and illustrated how such
measurements can dissect protein partnerships of a
whole protein family. This approach delivers a thermodynamic constant, the cooperativity factor, which allows
discriminating between competitive, additive and cooperative interaction modes of TF proteins. Our results suggest
that the protein interaction region of individual Sox HMG
domains encodes features enabling them to target specific
enhancers by teaming up with partner factors. By
contrast, the actual DNA recognition interface is very
similar and seems unable to explain the functional uniqueness of individual family members (2,26,30,33). A limitation of the approach presented here is its limited
throughput. However, high throughput methods such as
protein-binding microarrays (34) and HT-SELEX (3) have
not yet been adapted to identify composite DNA motifs of
heterodimers. Even if those methods can be adjusted to
multi-component systems, the development of computational tools that can accurately model cooperativity will
pose a significant challenge. Thus, we expect that our
method will complement high throughput efforts by
validating composite motifs and by providing quantitative
estimates of the physical cooperativity in TF-DNA
binding.
For Sox-Oct partnerships and probably many other TF
pairs, direct cooperativity is likely a major determinant for
the recruitment to functional-binding sites, and therefore a
major determinant of cell-type-specific biological function
(14,35). Thus, interrogating TF heterodimerization will
allow inferring coding principles for developmental enhancers and molecular mechanisms for selective and combinatorial enhancer recognition by TF proteins. We have
previously used such insights to re-engineer the endoderm differentiation factor Sox17 into an inducer of
pluripotency that speeds up stem-cell production (19).
The proof-of-concept that TFs can be optimized by
tweaking their heterodimerization and that their function can be rationally altered has broad application in
stem cell biology and tissue engineering for regenerative
medicine.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online:
Supplementary Table 1A–1C, Supplementary Figures 1
and 2.
ACKNOWLEDGEMENTS
The authors are grateful to Andrew Hutchins for critical
reading of the manuscript and suggestions and to Siew
Hua Choo for technical support. Author contributions:
C.K.L.N. and R.J.: conception and design, collection
and/or assembly of data, data analysis and interpretation,
manuscript writing, final approval of manuscript; S.P.:
derivation of cooperativity formula, data analysis and interpretation, manuscript writing; N.X.L., S.C.: collection
of data; P.K.: financial support, administrative support,
final approval of manuscript.
FUNDING
Funding for open access charge: Agency for Science,
Technology and Research (A*STAR) Singapore.
Conflict of interest statement. None declared.
REFERENCES
1. Bryne,J.C., Valen,E., Tang,M.H., Marstrand,T., Winther,O., da
Piedade,I., Krogh,A., Lenhard,B. and Sandelin,A. (2008)
JASPAR, the open access database of transcription factor-binding
profiles: new content and tools in the 2008 update. Nucleic Acids
Res., 36, D102–D106.
2. Badis,G., Berger,M.F., Philippakis,A.A., Talukder,S.,
Gehrke,A.R., Jaeger,S.A., Chan,E.T., Metzler,G., Vedenko,A.,
Chen,X. et al. (2009) Diversity and complexity in DNA
recognition by transcription factors. Science, 324, 1720–1723.
3. Jolma,A., Kivioja,T., Toivonen,J., Cheng,L., Wei,G., Enge,M.,
Taipale,M., Vaquerizas,J.M., Yan,J., Sillanpää,M.J. et al. (2010)
Multiplexed massively parallel SELEX for characterization of
human transcription factor binding specificities. Genome Res., 20,
861–873.
4. Biggin,M.D. (2011) Animal transcription networks as highly
connected, quantitative continua. Dev. Cell, 21, 611–626.
5. Kadonaga,J.T. (2004) Regulation of RNA polymerase II
transcription by sequence-specific DNA binding factors. Cell, 116,
247–257.
6. Mirny,L.A. (2011) Nucleosome-mediated cooperativity between
transcription factors. Proc. Natl. Acad. Sci. USA, 107,
22534–22539.
7. Glass,C.K. (1994) Differential recognition of target genes by
nuclear receptor monomers, dimers, and heterodimers. Endocr.
Rev., 15, 391–407.
8. Grove,C.A., De Masi,F., Barrasa,M.I., Newburger,D.E.,
Alkema,M.J., Bulyk,M.L. and Walhout,A.J. (2009) A
multiparameter network reveals extensive divergence between
C. elegans bHLH transcription factors. Cell, 138, 314–327.
9. Hai,T.W., Liu,F., Coukos,W.J. and Green,M.R. (1989)
Transcription factor ATF cDNA clones: an extensive family of
leucine zipper proteins able to selectively form DNA-binding
heterodimers. Genes Dev., 3, 2083–2090.
10. Garvie,C.W., Hagman,J. and Wolberger,C. (2001) Structural
studies of Ets-1/Pax5 complex formation on DNA. Mol. Cell, 8,
1267–1276.
11. Hollenhorst,P.C., Chandler,K.J., Poulsen,R.L., Johnson,W.E.,
Speck,N.A. and Graves,B.J. (2009) DNA specificity determinants
associate with distinct transcription factor functions. PLoS Genet.,
5, e1000778.
Nucleic Acids Research, 2012, Vol. 40, No. 11 4941
12. Kamachi,Y., Uchikawa,M., Tanouchi,A., Sekido,R. and
Kondoh,H. (2001) Pax6 and SOX2 form a co-DNA-binding
partner complex that regulates initiation of lens development.
Genes Dev., 15, 1272–1286.
13. Tanaka,S., Kamachi,Y., Tanouchi,A., Hamada,H., Jing,N. and
Kondoh,H. (2004) Interplay of SOX and POU factors in
regulation of the Nestin gene in neural primordial cells.
Mol. Cell. Biol., 24, 8834–8846.
14. Wilson,M. and Koopman,P. (2002) Matching SOX: partner
proteins and co-factors of the SOX family of transcriptional
regulators. Curr. Opin. Genet. Dev., 12, 441–446.
15. Kondoh,H. and Kamachi,Y. (2010) SOX-partner code for cell
specification: regulatory target selection and underlying molecular
mechanisms. Int. J. Biochem. Cell Biol., 42, 391–399.
16. Nasrin,N., Buggs,C., Kong,X.F., Carnazza,J., Goebl,M. and
Alexander-Bridges,M. (1991) DNA-binding properties of the
product of the testis-determining gene and a related protein.
Nature, 354, 317–320.
17. van de Wetering,M., Oosterwegel,M., van Norren,K. and
Clevers,H. (1993) Sox-4, an Sry-like HMG box protein, is a
transcriptional activator in lymphocytes. EMBO J., 12,
3847–3854.
18. Ryan,A.K. and Rosenfeld,M.G. (1997) POU domain family
values: flexibility, partnerships, and developmental codes.
Genes Dev., 11, 1207–1225.
19. Jauch,R., Aksoy,I., Hutchins,A.P., Ng,C.K., Tian,X.F., Chen,J.,
Palasingam,P., Robson,P., Stanton,L.W. and Kolatkar,P.R. (2011)
Conversion of Sox17 into a pluripotency reprogramming factor
by reengineering its association with Oct4 on DNA. Stem Cells,
29, 940–951.
20. Ambrosetti,D.C., Basilico,C. and Dailey,L. (1997) Synergistic
activation of the fibroblast growth factor 4 enhancer by Sox2 and
Oct-3 depends on protein-protein interactions facilitated by a
specific spatial arrangement of factor binding sites. Mol. Cell.
Biol., 17, 6321–6329.
21. Rodda,D.J., Chew,J.L., Lim,L.H., Loh,Y.H., Wang,B., Ng,H.H.
and Robson,P. (2005) Transcriptional regulation of nanog by
OCT4 and SOX2. J. Biol. Chem., 280, 24731–24737.
22. Nishimoto,M., Fukushima,A., Okuda,A. and Muramatsu,M.
(1999) The gene for the embryonic stem cell coactivator UTF1
carries a regulatory element which selectively interacts with a
complex composed of Oct-3/4 and Sox-2. Mol. Cell Biol., 19,
5453–5465.
23. Boyer,L.A., Lee,T.I., Cole,M.F., Johnstone,S.E., Levine,S.S.,
Zucker,J.P., Guenther,M.G., Kumar,R.M., Murray,H.L.,
Jenner,R.G. et al. (2005) Core transcriptional regulatory circuitry
in human embryonic stem cells. Cell, 122, 947–956.
24. Stefanovic,S., Abboud,N., Désilets,S., Nury,D., Cowan,C. and
Puceat,M. (2009) Interplay of Oct4 with Sox2 and Sox17: a
molecular switch from stem cell pluripotency to specifying a
cardiac fate. J. Cell Biol., 186, 665–673.
25. Ng,C.K., Palasingam,P., Venkatachalam,R., Baburajendran,N.,
Cheng,J., Jauch,R. and Kolatkar,P.R. (2008) Purification,
crystallization and preliminary X-ray diffraction analysis of the
HMG domain of Sox17 in complex with DNA. Acta Crystallogr.
Sect. F Struct. Biol. Cryst. Commun., 64, 1184–1187.
26. Palasingam,P., Jauch,R., Ng,C.K. and Kolatkar,P.R. (2009) The
structure of Sox17 bound to DNA reveals a conserved bending
topology but selective protein interaction platforms. J. Mol. Biol.,
388, 619–630.
27. BabuRajendran,N., Palasingam,P., Narasimhan,K., Sun,W.,
Prabhakar,S., Jauch,R. and Kolatkar,P.R. (2010) Structure of
Smad1 MH1/DNA complex reveals distinctive rearrangements of
BMP and TGF-{beta} effectors. Nucleic Acids Res., 38,
3477–3488.
28. Zhang,Y. (2008) I-TASSER server for protein 3D structure
prediction. BMC Bioinformatics, 9, 40.
29. Bowles,J., Schepers,G. and Koopman,P. (2000) Phylogeny of the
SOX family of developmental transcription factors based on
sequence and structural indicators. Dev. Biol., 227, 239–255.
30. Reményi,A., Lins,K., Nissen,L.J., Reinbold,R., Schöler,H.R. and
Wilmanns,M. (2003) Crystal structure of a POU/HMG/DNA
ternary complex suggests differential assembly of Oct4 and Sox2
on two enhancers. Genes Dev., 17, 2048–2059.
31. Chen,X., Xu,H., Yuan,P., Fang,F., Huss,M., Vega,V.B., Wong,E.,
Orlov,Y.L., Zhang,W., Jiang,J. et al. (2008) Integration of
external signaling pathways with the core transcriptional network
in embryonic stem cells. Cell, 133, 1106–1117.
32. Yuan,H., Corbi,N., Basilico,C. and Dailey,L. (1995)
Developmental-specific activity of the FGF-4 enhancer requires
the synergistic action of Sox2 and Oct-3. Genes Dev., 9,
2635–2645.
33. Jauch,R., Narasimhan,K., Ng,C.K. and Kolatkar,P.R. (2011)
Crystal structure of the Sox4 HMG/DNA complex suggests a
mechanism for the positional interdependence in DNA
recognition. Biochem. J., December 19 (doi:10.1042/BJ20111768;
epub ahead of print).
34. Berger,M.F., Badis,G., Gehrke,A.R., Talukder,S.,
Philippakis,A.A., Peña-Castillo,L., Alleyne,T.M., Mnaimneh,S.,
Botvinnik,O.B., Chan,E.T. et al. (2008) Variation in
homeodomain DNA binding revealed by high-resolution analysis
of sequence preferences. Cell, 133, 1266–1276.
35. Kamachi,Y., Uchikawa,M. and Kondoh,H. (2000) Pairing SOX
off: with partners in the regulation of embryonic development.
Trends Genet., 16, 182–187.